Past research on movements associated with speech production has been dominated by studies on small speaker samples, assuming essentially that individual differences in movement patterns are negligible. However, this assumption is inconsistent with published evidence. The broad goal of this project is to undertake detailed quantitative, statistically well- founded analyses of a unique, public-domain, large-sample speech production database incorporating point-parameterized representations of lingual, labial, and mandibular movements, recorded from the University of Wisconsin X-ray microbeam system in synchrony with the resulting acoustic sound pressure wave. Data from 57 normal speakers are represented, for a rich, uniform inventory of utterances and oral motor tasks, all described according to a common kinematic framework. Results from these analyses will be used to evaluate the general hypothesis that speakers differ in significant ways in the kinds and forms of movements used to produce perceptually equivalent speech behaviors. Specific aims include the development of speaker profiles, incorporating detailed models of individual strategies for changing speaking rate, clarity, and emphasis, complemented by information defining vocal tract size and speaker dialect; development of multi-variate time-series methods for describing patterns of articulatory coordination during sounds in varying contexts; and, description of the variability among speakers: choices of movement patterns whose acoustic effects are theoretically identical. An increased awareness of the range and distribution of speech production strategies among normals should prove useful for developing improved speech therapy protocols, and for achieving new understanding of the bases of speech disorders such as stuttering and the dysarthrias.